Testing machine learning of multimodal digital markers for early detection of cognitive impairment in Alzheimer's Disease rhoda
Bibliographic record
Abstract
BACKGROUND: Alzheimer's disease (AD) precision medicine will advance through the application of two key technological advances: 1) digital technologies that can more deeply characterize clinically relevant symptoms and 2) machine learning (ML) approaches the can classify subgroups with shared characteristics that could align with specific treatment plants. This study leverages a digital data collection platform for enhanced characterization and NetraAI, an artificial intelligence (AI) platform to analyze multimodal data to differentiate causal and non-causal subpopulations within a cohort and integrates a "No Call" system to exclude ambiguous data points. METHOD: We analyzed data from 98 Boston University Alzheimer's Disease Research Center participants and 453 variables derived from digital tasks administered over two months. Eight participants were clinically diagnosed as mild cognitive impairment. Digital measures included sleep metrics (57 measures), clinical scales (324 measures), and cognitive performance assessments (72 GoNoGo and Code Substitution measures). Of the 98 subjects, 81 were cognitively unimpaired and 17 transitioned to MCI during the course of study enrollment. RESULT: ). We examined 81 cognitively intact (e.g., non-transitioners; Class 0) and 17 MCI transitioners (Class 1) related to Go/No-Go and Code Substitution tasks. Go/No-Go Inter-Trial Intervals (ITI), REM sleep percentage, and maximum apnea duration were key predictors. Shorter, more stable ITI times (inter-trial intervals between tasks), higher REM sleep percentage, and shorter apnea durations were strongly correlated with non-transitioners. A 10-fold cross-validation yielded an average accuracy of 80.89%. CONCLUSION: Our findings present an ongoing effort on the potential of explainable AI to validate digital measures to identify those with MCI. While the current model effectively identifies prevalent non-transitioners, it remains limited in identifying prevalent transitioners. Future research will focus on refining model sensitivity and balancing classification performance.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.005 | 0.015 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".